Skip to content
Back to Blog
Insights

What does an AI-ready website need in Malaysia?

By Bernard Leong

Co-Founder, Geo One

17 November 2026
What does an AI-ready website need in Malaysia?

Short answer: an AI-ready website is one an assistant can reach, read and re-check. Three things decide that. The crawlers behind ChatGPT, Claude, Gemini and Perplexity aren't blocked at your firewall or content delivery network. Your core facts, what you do, where, for whom, at what price, under which licence, sit in plain text on the page rather than inside images, PDFs or a booking widget. And those same facts are marked up in structured data so a machine reads them the way a person does. Design, animation and brand photography matter to buyers. They have very little bearing on whether you get quoted.

This is written for owners and practice managers of Kuala Lumpur professional services firms with 2 to 50 staff: clinics, dental and veterinary practices, law and accounting firms, tuition centres, property agencies. Disclosure: it comes from Geo One, a Generative Engine Optimisation (GEO) and Answer Engine Optimisation (AEO) agency in Kuala Lumpur.

Why build for assistants at all?

Because the traffic is real and it behaves differently. WebFX reported that generative AI referral traffic grew 796% between January 2024 and December 2024, based on analysis of 2.3 billion site sessions, and that AI-referred visitors converted at roughly 1.2 times the rate of organic search visitors. That second number is the interesting one for a KL firm. A person who arrives after an assistant has already described what you do has done half their shortlisting before they land.

There's a governance angle too. Malaysia's Ministry of Science, Technology and Innovation published the National Guidelines on AI Governance and Ethics on 20 September 2024, voluntary, and built on principles including transparency and accountability. Nothing in it obliges you to touch your website. It does set the direction of travel: stating plainly who you are and what you claim is becoming the baseline, not a differentiator.

Can the assistants actually fetch your pages?

Start here, because the rest of the work has less impact if the answer is no. Different assistants send different crawlers, and a single rule in a bot management panel can allow one while challenging another. We see this often on Malaysian SME sites where a developer set up the content delivery network two or three years ago and has since moved on.

Check the current documentation rather than trusting a setting from memory. OpenAI describes how GPTBot and its other agents identify themselves in the OpenAI platform documentation, and if you sit behind Cloudflare, the bot and crawler controls are set out in the Cloudflare developer docs. Anthropic and Perplexity publish their own crawler notes, and each operates more than one agent, so blanket rules rarely do what the person writing them intended. We've written a longer walkthrough of the Cloudflare AI crawler settings that quietly block assistants if you want to check your own setup.

One practical note for firms with a members area or online booking: if your service descriptions, fees or practitioner credentials sit behind a login, they're not readable. A public summary page that states the same facts solves it without exposing anything you'd rather keep private.

The same applies to the local habit of keeping prices entirely on WhatsApp. "WhatsApp us for a quote" is a reasonable sales tactic and a dead end for an assistant, there is nothing on the page to lift. A starting price or a range gives it something to work with, and you can still settle the real number in the chat.

Does your site state the facts plainly enough to be re-used?

An assistant is trying to answer a question with something it can stand behind. Vague marketing copy gives it nothing to work with. "We deliver client-focused solutions" cannot be re-stated as a fact. "We handle SST registration and monthly filing for Klang Valley trading companies, from RM600 a month" can.

So write the specifics down. Services, with the words your buyers actually use. Who you serve and who you don't. Locations and the areas you cover. Languages your staff work in, which matters more than most KL firms realise when a buyer asks in Bahasa Malaysia or Mandarin. Opening hours. A real phone number.

And name the body behind your credentials, not just the credential. The Malaysian Bar for law firms, the Malaysian Medical Council or Malaysian Dental Council for clinics, the Malaysian Institute of Accountants for accounting practices, the Board of Valuers, Appraisers, Estate Agents and Property Managers for agencies. "Registered practitioner" with no issuing body is not a checkable fact, and an assistant that cannot check a claim tends to leave it out rather than repeat it.

Then mark it up. Structured data is a standard vocabulary for describing a business in code, maintained at Schema.org, and it's how a machine knows that a string of digits is a phone number rather than a reference code. LocalBusiness, Organization, Service, Person and FAQPage cover most professional services firms. If you want to see whether yours is readable today, there's a simple structured data test you can run on your own site.

The last piece is consistency, and in Malaysia it usually fails the same way. The business is registered with the Companies Commission of Malaysia (SSM) under one name, trades under a shorter one, and the Google Business Profile carries a third with the branch suburb appended. Directory listings then pick whichever they found first. Every version is defensible on its own. To an assistant, they look like three similar businesses at nearby addresses, and none of them is clearly the one being asked about. Pick the trading name, use it identically everywhere including the address format, and keep the registered entity name in one visible place, about page or footer, where it can be matched rather than guessed. Matching all this up is dull work. It's also the cheapest improvement available to most firms.

What won't an AI-ready website fix on its own?

It won't make you the named answer. A site is the thing an assistant verifies against. It is not the thing that puts your name in the sentence. What tends to do that is other credible sources describing you the same way.

Our own numbers make the point. In an AI visibility check we ran in late 2025, we looked at 85 answers from ChatGPT, Claude, DeepSeek, Gemini and Perplexity. Geo One was named in 7 of them, for "top AI visibility agencies in Kuala Lumpur" but not for "best generative engine optimisation agency in Kuala Lumpur". Same company, same website, same structured data, two different results. The gap sits in third-party coverage, not in our markup.

We'd rather say that plainly than sell you a rebuild as a complete answer. A good site makes you citable. Being cited is a separate job, and no one can promise a particular assistant will name you on a particular day.

Rebuild, or fix what you have?

Most KL firms don't need a new website. They need the pages they already have to say more and hide less. A typical first pass is a services page per service instead of one combined page, an about page that states registration and credentials, an FAQ written in the words buyers type, structured data across the lot, and the crawler rules checked.

If the existing site renders everything through JavaScript and serves almost no text in the initial HTML, a rebuild is sometimes the shorter route. That's a judgement call made after looking, not before.

Geo One works from RM500 to RM2,000 a month, covering scanning, schema and entity fixes, content and citation work, with AI agents running the scanning and reporting weekly. The full breakdown sits on our pricing page. If you'd rather just know whether assistants can read your site today, send us the URL by WhatsApp or the enquiry form and we'll tell you what we find.

Frequently asked questions

Why build for assistants at all?

Because the traffic is real and it behaves differently. WebFX analysed 2.3 billion site sessions between January 2024 and December 2025 and [found generative AI referral traffic grew 796% over that period](https://www.webfx.com/blog/seo/gen-ai-search-trends/), with AI-referred visitors converting at roughly 1.2 times the rate of organic search visitors. That second number is the interesting one for a KL firm. A person who arrives after an assistant has already described what you do has done half their shortlisting before they land.

Can the assistants actually fetch your pages?

Start here, because the rest of the work has less impact if the answer is no. Different assistants send different crawlers, and a single rule in a bot management panel can allow one while challenging another. We see this often on Malaysian SME sites where a developer set up the content delivery network two or three years ago and has since moved on.

Does your site state the facts plainly enough to be re-used?

An assistant is trying to answer a question with something it can stand behind. Vague marketing copy gives it nothing to work with. "We deliver client-focused solutions" cannot be re-stated as a fact. "We handle SST registration and monthly filing for Klang Valley trading companies, from RM600 a month" can.

What won't an AI-ready website fix on its own?

It won't make you the named answer. A site is the thing an assistant verifies against. It is not the thing that puts your name in the sentence. What tends to do that is other credible sources describing you the same way.

Bernard Leong

Co-Founder, Geo One

Nearly 20 years across energy, capital strategy and applied AI, including large-scale operational data at BP. Founded SkillsMe and Cryptrain.

More about the team

Check Your AI Visibility

See how AI platforms currently view your business with a free scan.

Free AI Scan